Identification and Selective Degradation of Neopeptide-Containing Truncated Mutant Proteins in the Tumors with High Microsatellite Instability

Activating mutations in oncogenes and inactivating mutations in tumor suppressor genes are hallmarks of human cancers (1). The common inactivation pathways of tumor suppressor genes are deletion of one allele and inactivating mutations of the other allele (2). Of these pathways, nonsense and frameshift mutations are the most critical inactivating mutations (3). These 2 types of mutations inactivate genes by introducing premature termination codons (PTC) in mRNA. If mRNAs bearing PTCs are not degraded and allowed to be translated normally, then genes with nonsense mutations are expected to generate truncated proteins lacking neopeptides, whereas genes with frameshift mutations are expected to generate truncated proteins containing neopeptides. Generally, most abnormal PTC-containing mRNAs are actively degraded by nonsense-mediated mRNA decay (NMD), thus avoiding the potentially deleterious effects associated with the production of truncated proteins (4, 5). NMD is mediated through the recognition of PTC-containing mRNAs, which are recognized by their position relative to the last exon–exon junction. Mammalian transcripts that contain PTCs more than 50 to 55 nucleotides (nt) upstream of the last exon–exon junction are degraded by NMD, which ensures the degradation of most PTC-containing mRNAs. However, PTCs located within 50 to 55 nt or downstream of the last exon–exon junction are not recognized by NMD and can potentially lead to the generation of mutant proteins (6, 7).

A subset of colorectal tumors exhibits length alterations in several coding and noncoding microsatellites, a molecular phenotype termed high microsatellite instability (MSI-H; refs. 8, 9). The length alterations in microsatellites of the coding region [coding mononucleotide repeats (cMNRs)] result in frameshift mutations in the affected genes, and these mutations are believed to contribute to tumor development and progression (10). Although many reports indicated that numerous genes are frequently mutated in their cMNRs in MSI-H cancers, few of these genes have been reported to express their mutant gene products in MSI-H cancers (11–13). We searched for genes containing cMNRs in the last exon and analyzed the frequency of mutations in these genes. We addressed whether genes with frameshift mutations generate truncated mutant proteins. We showed that mutant proteins are actively translated from genes containing mutations in cMNRs in the last exon but are rarely detected because these endogenous truncated proteins containing neopeptides are extensively degraded by the ubiquitin–proteasome system.

For the mutation analysis, 12 cell lines were used. DLD1, HCT116, HCT-8, LOVO, LS174T, RKO, SNUC2A, SNUC2B, SNUC4, and SNU407 cells are MSI-H colorectal carcinoma cell lines, whereas WiDr and HeLa cells are microsatellite-stable (MSS) cell lines, as determined by previous studies (14, 15). Cells were grown in RPMI, minimum essential medium, and Dulbecco's modified Eagle medium supplemented with 10% FBS (Life Technologies), 1% penicillin, and streptomycin at 37°C in 5% CO₂. About tissue samples, 19 specimens confirmed as MSI-H colorectal carcinomas using BAT25, BAT26, D5S346, D17S25, and D2S123 were included in this study. Some of the fresh specimens were obtained from the Liver Cancer Specimen Bank of the National Research Resource Bank Program of the Korea Science and Engineering Foundation of the Ministry of Science and Technology. Authorization for the use of the tissues for research was obtained from the Institutional Review Board. Conventional pathologic parameters were examined without prior knowledge of the molecular data (Table 1).

Genomic DNA and cDNA preparation, analysis of MSI, and identification of target gene mutations were conducted using a PCR-based assay as described previously (16).

The primers for semiquantitative reverse transcription (RT)-PCR and quantitative reverse transcription PCR (qRT-PCR) were designed using Primer 3 database (http://frodo.wi.mit.edu/primer3/). All RNAs were isolated from cells using TRIzol (Invitrogen). Reverse transcription was conducted using M-MLV reverse transcriptase (Invitrogen). For RT-PCR, the reaction was conducted using AmpliTaq Gold 360 DNA Polymerase (Applied Biosystems). For qRT-PCR, the reaction was conducted using the ABI PRISM 7500 Sequence Detector (Applied Biosystems) and SYBR Premix Ex Taq II (TaKaRa). The amount of target mRNA was normalized to that of GAPDH or EGFP mRNA [derived from the enhanced GFP (EGFP)-expressing control vector]. The sequences of the primers used are listed in Supplementary Table S1.

cDNA expression vectors for TTK [cTTK(WT)], TCF7L2 [cTCF4L2(WT)], and MARCKS [cMARCKS(WT)] were constructed by cloning the respective wild-type (WT) genes into pcDNA3.1 vectors containing a FLAG tag via amplification of their coding regions using the cDNAs derived from HeLa and WiDr cells. For the genomic DNA form of the MARCKS expression vector [gMARCKS(WT)], all exons and introns between the exons of MARCKS were cloned into pcDNA3.1 vectors containing a FLAG tag. To generate mutant protein expression vectors for TTK [cTTK(−2)], TCF7L2 [cTCF7L2(−1)], and MARCKS [cMARCKS(−2), gMARCKS(−2)], deletion mutagenesis was conducted. All transfection experiments were carried out using Lipofectamine 2000 (Invitrogen), and a CMV10-EGFP vector was used to confirm the transfection efficiency. The primers used for cloning are listed in Supplementary Table S1. The siRNAs against TTK, TCF7L2, and MARCKS used in this study (Bioneer) were also transfected into cells using Lipofectamine.

Whole lysates from cells were prepared using passive lysis buffer (Promega). Membranes were incubated with primary antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Trevigen), FLAG (Sigma-Aldrich), TTK (Abnova), TCF7L2 (Cell Signaling Technology), and MARCKS (Santa Cruz Biotechnology) for 1 hour at room temperature. Antibodies against the neopeptide sequences of mutant MARCKS(−2) were generated in rabbits. All antibody generation procedures were conducted according to the manufacturer's manuals (Young In Frontier). More detailed information on the antibody generation is included on the website (http://www.molpathol.org/).

HeLa cells were washed 3 times with ice-cold PBS containing 100 μg/mL cycloheximide, collected, and then lysed in lysis buffer [15 mmol/L Tris–Cl (pH 7.4), 3 mmol/L MgCl₂, 10 mmol/L NaCl, 0.5% Triton X-100, 100 μg/mL cycloheximide, 1 mg/mL heparin, and 200 U RNasin (Promega)]. Nuclei and debris were removed by centrifugation at 12,000 × g for 2 minutes. One milliliter of each cytoplasmic lysate was layered onto an 10% to 50% sucrose gradient and centrifuged at 4°C (39,000 rpm) for 2 hours. Sixteen fractions were collected with concomitant measurements of absorbance at 254 nm by using a fraction collection system. RNA was extracted with TRIzol and analyzed by RT-PCR.

Subcellular localization of mutant MARCKS was analyzed by immunofluorescence staining. The cells attached to glass coverslips were rinsed with PBS followed by fixation and permeabilization with ice-cold methanol for 10 minutes at −20°C. Upon the removal of methanol, cells were again rinsed. Nonspecific sites were blocked with 2% bovine serum albumin for 1 hour. After blocking, the medium was replaced with the respective primary antibodies, and cells were incubated overnight. Cells were then washed and incubated for 1 hour with the appropriate fluorescently labeled secondary antibodies. For double-labeling experiments, cells were simultaneously incubated with the primary and secondary antibodies. Anti-FLAG M2-FITC (Sigma; F4049) and anti-γ-tubulin (Sigma; T5192) were used in this experiment. All images were obtained using an LSM700 confocal microscope (Carl Zeiss).

Immune complexes of wild-type and mutant proteins were collected by gently rocking 1 mg of total proteins on an orbital shaker with prewashed anti-FLAG M2-agarose affinity gel (Sigma-Aldrich) at 4°C. The immune complexes bound to the affinity gel were washed and then boiled with a 100 mmol/L Tris–HCl–1% SDS solution to elute the complexes. Western blotting was conducted using FLAG and HA (Santa Cruz Biotechnology) antibodies. The relative density of each lane was quantified by ImageJ (NIH, Bethesda, MD) software.

We searched for human genes containing cMNRs longer than 9 nt in SelTarbase (http://www.seltarbase.org), a database providing comprehensive information about human mononucleotide microsatellite mutations and genes containing cMNRs. In total, 447 genes satisfying these criteria were obtained from SelTarbase. To confirm that these 447 genes include cMNRs longer than 9 nt, we analyzed the entire genes using bioinformatic tools such as Vector NTI (Invitrogen), the Human BLAST database (http://genome.ucsc.edu/), and the National Center for Biotechnology Information gene database. With this approach, 302 of 447 genes were confirmed to have cMNRs longer than 9 nt. On the basis of the NMD-irrelevant condition (mRNAs containing PTCs within 50–55 nt of the last exon junction), 302 genes were manually analyzed using Vector NTI software. When 1- or 2-bp deletions/insertions were detected in the cMNR region, genes that acquired abnormal stop codons distal to a site 50 to 55 nt from the last exon junction complex were selected. The number of finally selected genes was 66. Among them, 15 genes had previously been reported to have mutations in their cMNRs in MSI-H cancers (Fig. 1A; http://www.sanger.ac.uk/genetics/CGP; refs. 11, 12). Most of the abnormal stop codons were PTCs; however, abnormal stop codons after normal stop codons (readthrough) were also found in some genes with cMNR mutations. Many of the 66 genes were related to biologically critical reactions, such as apoptosis, cell-cycle regulation, cell proliferation, angiogenesis, and intracellular signaling (Supplementary Table S2).

We randomly selected 35 of 66 genes to analyze cMNR mutations by conducting an isotope PCR-based assay and sequencing. We used 10 MSI-H colon cancer cell lines for the mutation search. A MSS colon cancer cell line (WiDr) and the HeLa cell line were used as controls. We found frequent frameshift mutations in the cMNRs of 29 genes in the 10 MSI-H cell lines. The mutation profiles and status (homozygous vs. heterozygous) are summarized in Table 2. To validate the PCR-based mutation analysis, we conducted sequence analysis of several genes in the cell lines with homozygous mutations (Fig. 1B). All 10 MSI-H cancer cell lines had mutations in more than 4 of 35 genes examined, and 19 mutated genes were found in SNUC2A and SNUC2B cells (Fig. 1C). The mutation frequency of the 29 genes varied from 10% to 90% in the 10 MSI-H cell lines (Fig. 1D). We selected TTK, TCF7L2, and MARCKS and conducted PCR-based mutation analysis in 19 MSI-H colon cancer tissues. All 3 genes displayed frequent and varying mutation incidences ranging from approximately 35% to 60% (Fig. 1E). Information about the tissue samples used in this study is provided in Table 1.

For the mutant protein expression analysis, we firstly selected 7 genes (TTK, TCF7L2, MARCKS, ASTE1, INO80E, CYHR1, and EBPL) displaying frameshift mutations in MSI-H cancer cell lines according to the availability of antibodies, presence of homozygous mutations, and cancer relevance. Before Western blot analysis, the mRNA expression of each gene was measured by qRT-PCR and semiquantitative RT-PCR analysis. From these experiments, we sought to confirm that the mRNAs from mutated genes are not decayed by NMD or downregulated by other genetic events, such as deletion or methylation. The mRNA expression of the selected genes was similarly quantified in the cell lines with homozygous mutations and compared with their levels in cells without mutations or those with heterozygous mutations in one allele (Supplementary Fig. S1). SNUC4 cells, a MSI-H cell line with homozygous mutations in TTK (+1/−2), TCF7L2 (−1/−1), and MARCKS (−1/−1), exhibited relatively higher TTK, TCF7L2, and MARCKS mRNA levels than the other cell lines (Fig. 2A–C). These results indicate that mRNAs containing PTCs in the last exon are intact in cells irrespective of the mutation status of genes.

We tested all the antibodies against the selected 7 genes by Western blotting, and found that only antibodies against TTK, TCF7L2, and MARCKS showed excellent sensitivity and specificity for the experiments to follow. Then, we analyzed the expression of endogenous mutant TTK, TCF7L2, and MARCKS in 9 cell lines (7 MSI-H colon cancer cell lines, a MSS colon cancer cell line, and the HeLa cell line) by Western blotting. There was a minimal size difference between the mutant and normal proteins generated from TTK, whereas considerable size differences were observed in the normal and mutant proteins generated from TCF7L2 and MARCKS (Fig. 2D). We expected that if mutant proteins are expressed, then heterozygous mutation will cause quantitative and qualitative differences in the proteins and homozygous mutations will cause qualitative differences in the proteins because no wild-type proteins can be generated. To confirm our hypothesis, we quantified the normal TTK, TCF7L2, and MARCKS levels in each cell line based on their mutation status. Because of the heterozygous mRNA expression level in each cell line, we calculated the amount of normal protein by normalizing the protein level to the mRNA level, and then mean values were obtained depending on the mutation status. The result revealed that the normal protein level decreased according to the mutation status (wild-type > heterozygous mutation > homozygous mutations; Supplementary Fig. S2). For each mutant protein, cell lines with no mutations (4 cell lines) or heterozygous mutations (4 cell lines) in TTK exhibited a positive band at 90 kDa. The size difference between wild-type and mutant TTK was 0.3 kDa, and thus, normal and mutant TTK cannot be distinguished by Western blotting in these cell lines. Therefore, SNUC4 cells represented the proper model for confirming whether mutant TTK is expressed because these cells have homozygous mutations in TTK. Although SNUC4 cells exhibited relatively higher TTK mRNA expression than the other cell lines used, only faint mutant TTK expression was detected (Fig. 2E). The size difference between wild-type and mutant TCF7L2 was approximately 24 kDa. We detected faint bands only in cell lines with −1/−1 homozygous or −1/w heterozygous deletions (LS174T, SNUC4, and SNU407) compared with the patterns in cell lines with no mutations (Fig. 2E). The size difference between wild-type and mutant MARCKS was approximately 35 kDa. A complete loss of both normal and mutant MARCKS was observed in the 4 cell lines (LoVo, LS174T, SNUC4, and SNU407) with homozygous mutations (Fig. 2E). On the basis of these results, we suspected that either translation of the mutant mRNAs is repressed or mutant proteins are normally generated but extensively degraded through protein degradation pathways.

To evaluate the efficiency of the translation of mRNAs containing PTCs in the last exon, we generated the gMARCKS(WT) vector, a genomic DNA vector construct composed of 2 exons and 1 intron of MARCKS. We also generated the gMARCKS(−2) vector, a genomic DNA mutant MARCKS vector missing 2 adenine residues in the cMNR region (A11) of the last exon. We examined the translation of mutant mRNAs from the gMARCKS(−2) vector by analyzing the distribution of polysomes, using the gMARCKS(WT) vector as a control. Puromycin treatment was used to mimic a condition in which translation is repressed. The results revealed that wild-type MARCKS mRNA from HeLa cells transfected with the gMARCKS(WT) vector was mostly present in the polysome fractions (right shifted), and the GAPDH mRNA distribution was similar, which indicated active translation (Fig. 3A). Mutant MARCKS mRNA-bearing PTCs from HeLa cells transfected with the gMARCKS(−2) vector exhibited a similar pattern as the wild-type MARCKS and endogenous GAPDH mRNAs (Fig. 3B). The normal translation of both wild-type and mutant MARCKS mRNAs was confirmed by the left shift in the banding pattern after puromycin treatment, indicating translational repression (Fig. 3C and D). On the basis of this polysome analysis, we concluded that both wild-type and mutant MARCKS mRNAs are actively translated.

According to polysome analysis, we suspected that mutant proteins are generated but mostly degraded through proteolytic pathways. To determine which proteolytic pathways are responsible for degrading mutant proteins, we designed a rescue assay using several proteolytic inhibitors of 3 major protein degradation pathways: proteasome-, autophagy-, and lysosome-mediated degradation (17, 18). Bafilomycin A1 was used to block lysosomal degradation, and 3-methyladenine was used to inhibit autophagy-mediated degradation. Lactacystin and MG132 were used to block the proteasomal pathway. For this experiment, we also generated mutant cDNA vectors of TTK [cTTK(−2)], TCF7L2 [cTCF7L2(−1)], and MARCKS [cMARCKS(−2)] to generate control proteins (Fig. 4A). The rescue assays for mutant TTK and TCF7L2 were conducted using SNUC4 cells, and SNUC2B cells were used for mutant MARCKS. The result showed that the levels of endogenous mutant TTK were significantly increased (approximately 2.5-fold) after proteasome inhibition, but no increment was observed when autophagy- or lysosome-mediated degradation was blocked. Expression of the TCF7L2 mutant was also greatly increased (∼4-fold) only after proteasome inhibition in SNUC4 cells. For the detection of mutant MARCKS, we generated a more sensitive and specific antibody against the neopeptide sequences in the C-terminal region of mutant MARCKS (Supplementary Fig. S3A). This antibody more specifically and sensitively detected mutant MARCKS than the antibody recognizing the N-terminal region of MARCKS (Supplementary Fig. S3B). We conducted a mutant MARCKS rescue assay using the generated antibody. Interestingly, the expression of mutant MARCKS was increased by more than 100-fold when proteasome degradation was blocked in SNUC2B cells (Fig. 4B). We clarified these facts by using RNA interference (RNAi) against TTK, TCF7L2, and MARCKS. In SNUC4 cells treated with siRNA against TTK, the proteasome inhibition-induced increase in mutant TTK expression was reduced to approximately 45% of that in MG132-treated SNUC4 cells, confirming that these bands represent TTK. To confirm whether MG132 treatment specifically rescues mutant TTK, we conducted the same experiment using HeLa cells, which only express wild-type TTK. The results revealed that MG132 treatment specifically affects mutant TTK protein (Supplementary Fig. S4A). As observed for mutant TTK, we showed that the proteasome inhibition–induced increases in mutant MARCKS and TCF7L2 expression were reduced by approximately 50% upon siRNA treatment. However, no changes were observed in HeLa cells after proteasome inhibition or siRNA treatment (Supplementary Fig. S4B and S4C). We examined the stability of endogenous mutant TTK in a time course experiment using SNUC4 (+1/−2) and HeLa cells (wild-type). The expression of mutant TTK was dramatically decreased in the absence of MG132, which confirms that endogenous mutant TTK is rapidly degraded by the proteasome system (Fig. 4C). The expression of wild-type TTK was stable during the experiment (Fig. 4D).

In addition to the rescue assays for endogenous mutant proteins, we further showed that mutant TTK, TCF7L2, and MARCKS are more unstable than their wild-type counterparts and rapidly degraded by the proteasome system when overexpressed in vitro. As expected, the mutant TTK, TCF7L2, and MARCKS levels were approximately 19%, 11%, and 13%, respectively, of their wild-type levels. Inhibition of proteasome degradation by MG132 increased the expression of the mutants by approximately 3-fold and mRNA expression level from the vector constructs was similar (Supplementary Fig. S5A–S5C). To confirm the involvement of the proteasome system, we conducted an ubiquitination assay using the mutant and wild-type cDNA constructs of TTK, TCF7L2, and MARCKS. Consequently, the 3 mutant proteins were more heavily ubiquitinated (more than 2-fold) than the wild-type proteins in the presence of MG132 (Fig. 5A–C). On the basis of the ubiquitination assay, we further examined the localization of mutant proteins using immunofluorescence microscopy. According to previous studies, we hypothesized that if mutant proteins are specifically localized around centrosomes, which recruit the proteasomal machinery, then this further confirms that mutant proteins are actively degraded (19, 20). We conducted immunofluorescence staining using antibodies against FLAG and γ-tubulin, the latter of which was used as a centrosome marker. We chose mutant MARCKS constructs for the experiment. Under confocal microscopic examination, proteasome inhibition resulted in centrosomal expansion in cells transfected with both wild-type and mutant MARCKS constructs. However, colocalization with γ-tubulin was detected only in cells transfected with the mutant construct, indicating that mutant proteins are actively recruited to centrosomes after proteasome inhibition (Fig. 5D). Taken together, our results suggest that both endogenous and synthetic mutant proteins are more rapidly degraded than wild-type proteins, and this process is mediated via the ubiquitin–proteasome pathway.

All genes containing frameshift mutations in cMNRs in the last exon are expected to generate truncated proteins with variable lengths of neopeptides. After we showed that the endogenous neopeptide-containing truncated mutant proteins were mostly degraded in the proteasome, we hypothesized that neopeptides and/or protein truncation might enhance mutant protein degradation. To validate our hypothesis, we constructed another protein expression vector by using the MARCKS genomic DNA construct. The mutant protein produced from the gMARCKS(−2) vector was composed of 183 amino acids, 27 of which comprised neopeptides in the C-terminal region. We generated an additional vector construct by introducing an abnormal stop codon at amino acid position 183 [gMARCKS(p183)], which leads to generation of truncated mutant proteins lacking neopeptides (Fig. 6A). We evaluated MARCKS mRNA and protein expression levels before and after MG132 treatment. The expression of mRNAs from the cells transfected with the gMARCKS(−2) and gMARCKS(p183) vectors was similar to that in the cells transfected with the wild-type MARCKS vector [gMARCKS(WT); Fig. 6B, right]. Comparing the protein expression level, we found that the expression of synthetic wild-type MARCKS protein was higher than that of synthetic mutant MARCKS proteins irrespective of the presence of neopeptides. Interestingly, cells transfected with gMARCKS(p183) exhibited nearly 2-fold higher mutant protein levels than cells transfected with gMARCKS(−2) in the absence of MG132. After proteasome inhibition, the levels of neopeptide-containing mutant MARCKS were almost doubled, whereas those of the mutant lacking neopeptides were only increased slightly (Fig. 6B, left). The ubiquitination assay indicated that the mutant MARCKS containing neopeptides is more heavily ubiquitinated than the mutant MARCKS lacking neopeptides, indicating that neopeptides are primarily responsible for the degradation of mutant MARCKS (Fig. 6C). As the expression of mutant MARCKS remained lower than that of wild-type MARCKS even after proteasome inhibition, we suspected that other factors might be involved in the low expression of mutant MARCKS. Thus, we hypothesized that the changes in mutant protein levels could be related to increases in insolubility, which contributes to the formation of insoluble bodies in cells. To show whether the formation of insoluble bodies reduces mutant protein expression, we fractionated the cell lysates into Triton X-100–soluble and Triton X-100–insoluble fractions. GAPDH and γ-tubulin were used as Triton X-100–soluble markers. Surprisingly, significant levels of both mutant MARCKS proteins were detected in the Triton X-100–insoluble fraction, indicating that the low expression of mutant MARCKS after proteasome inhibition was due to the increased insolubility of the mutant proteins and subsequent formation of insoluble bodies (Fig. 6D). These findings indicate that truncated mutant proteins containing neopeptides are rarely detected because of extensive degradation and increased insolubility (Fig. 6E).

Mutations are hallmarks of diseases, especially cancers (21, 22). Mutations contribute to cancers via gain-of-function mutations of oncogenes and loss-of-function mutations in tumor suppressor genes (23). The generation and fate of mutant proteins produced from mutated genes vary according to the gene and type of mutation. Mutations in the oncogenes such as KIT and β-catenin result in the overexpression of activated KIT and β-catenin in many human cancers (24, 25). Overexpression of the wild-type oncoproteins HER2 and cMYC due to gene amplification is also well established (26, 27). About tumor suppressor genes, downregulation of wild-type proteins is common due to inactivation of both alleles (28). The inactivation mechanisms are deletion, methylation inactivation, and inactivating mutation (29, 30). No functional proteins can be expressed from the tumor suppressor genes with homozygous deletions or those with deletion of one allele and methylation inactivation of the other allele. In cases of inactivating mutations, however, expression of the functional protein might be variable according to the genes and type of mutations. For example, p53 ablation via the deletion of one allele and a point mutation in the other allele is very frequent in cancers, and mutant p53 overexpression due to the dysregulation of ubiquitination has been reported (31–33). To this point, the protein expression of many other cancer-related genes with inactivating mutations remains unknown.

Nonsense and frameshift mutations are important inactivating mutations in the development of genetic diseases and human cancers. In particular, in human MSI-H cancers, frequent frameshift mutations have been reported in many genes (8, 10, 12). When frameshift mutations induce randomized nucleotide arrangement after insertion or deletion sites, the probability of stop codon generation is approximately 3/64 (3 stop codons for every 64 codons), and therefore, the average neopeptide is expected to be approximately 20 amino acids in length. In 29 genes, in which we showed frameshift mutations, the mean length of neopeptides induced by the 1-bp deletion and 2-bp deletion mutations was 22.7 ± 17, and 28.5 ± 52.8, respectively (Supplementary Fig. S6A–S6C). Because we confirmed that mRNAs derived from mutated genes were intact in the cytoplasm, the NMD-irrelevant mRNA model is essential to determine if truncated mutant proteins with neopeptides are expressed in tumors. Previously, several studies have attempted to detect truncated mutant proteins derived from PTC-carrying mRNAs and clarify the roles of these mutant proteins in diseases such as cancers and genetic diseases (13, 34–39). Thus far, most of the studies have shown the existence of mutant proteins at the DNA or mRNA level (34, 37). Some studies suggested the existence of mutant proteins by in vitro overexpression experiments, but the constructs used in the previous studies had PTCs in specific positions of the C-terminal region, which represents nonsense mutations (38–40). In most cancers, insertion/deletion mutations usually lead to frameshift mutations that are much more frequent and deleterious than nonsense mutations. Therefore, the identification of endogenous truncated proteins containing neopeptides from genes with frameshift mutations is more important, but this has not been studied at the protein level. We herein validated for the first time the expression of the truncated mutant proteins derived from the genes with frameshift mutations in protein level.

For the validation of endogenous truncated mutant protein expression, we chose TTK, TCF7L2, and MARCKS. The physiologic roles of TTK, TCF7L2, and MARCKS have been studied in several cancers, and these proteins have significant relevance to cancer progression (41–44). We originally expected that significant amounts of mutant TTK might be expressed because this mutant protein has minor changes and a readthrough stop codon. To our surprise, endogenous mutant TTK was barely detected by Western blotting despite the relatively high mRNA expression level of TTK in SNUC4 cells, which have homozygous mutations (+1/−2) in TTK. Mutant TCF7L2 was also barely expressed, and mutant MARCKS was not expressed in the tumor cells with homozygous mutations. After proteasome inhibition by MG132 treatment, large amounts of mutant TTK, TCF7L2, and MARCKS were detected, and we observed these dramatic increases only in the cells expressing the mutant proteins. These findings clearly indicated that the rare expression of mutant proteins of these 3 genes is mostly due to enhanced degradation. In addition to showing the dramatic increment of mutant protein expression after proteasome inhibition, we also showed the selective and heavy ubiquitination of the mutant proteins. Moreover, the colocalization of mutant MARCKS with centrosomes in the presence of MG132 suggests that mutant MARCKS is actively degraded via the proteasomal machinery, which is assembled around centrosomes. By generating 2 different truncated mutant MARCKS expression vectors that contained or lacked neopeptides, we showed that neopeptide-containing mutant MARCKS proteins are more extensively degraded. Furthermore, we found that the neopeptide-lacking mutant proteins were relatively stable compared with the neopeptide-containing mutant proteins, and the truncated mutant displayed increased insolubility irrespective of the presence of neopeptides. Our data explain why the mutant proteins are barely detected, and reveal that tumor cells are protected from potentially harmful mutant proteins by both NMD- and the ubiquitin-mediated protein degradation mechanism. In addition to these results, the functional relevance of scant amount of detected neopeptide-containing mutant proteins in the tumor progression should be studied in the future.

Our findings suggest that the degraded mutant proteins contribute to the formation of tumor-specific antigens and these antigens are useful targets for immunotherapy. Intense peritumoral and intratumoral lymphocytic infiltration, and its association with favorable prognosis have been reported as the characteristics of MSI-H colon cancers (8, 45, 46). It is also very clear that the amount of intracellular mutant proteins, a substrate for tumor antigen, is closely related to the effective tumor antigen formation (47). Our results, protein translation from the NMD-irrelevant mutant mRNAs and the generalized degradation of neopeptide-containing mutant truncated proteins, provide novel insights that (i) the intracellular amount of mutant proteins are scant, but (ii) the degradation of these neopeptide-containing mutant proteins by proteasome system is directly related to the antigen-processing and presentation by MHC class I, therefore expected to be effective tumor antigen formation. When we analyzed 19 MSI-H colon cancer tissues, we found a significant relationship between the intensity of peritumoral reaction and the mutation status of 3 genes [TTK (P = 0.01), TCF7L2 (P = 0.46), and MARCKS (P = 0.002); Table 1]. These findings suggest that the degraded mutant proteins might be related to the regional immune responses of the tumor. A large-scale correlation study and a study on the immunostimulatory function of the neopeptides will be necessary to determine the roles of mutant proteins in tumor antigen formation, and mutations in other genes containing cMNRs in the last exon also should be evaluated.

No potential conflicts of interest were disclosed.

Conception and design: W.K. Kim, K.T. You, H. Kim

Development of methodology: W.K. Kim, K.T. You, H. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.K. Kim, M. Park, H.K. Kim, H. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W.K. Kim, M. Park, N. Shin, H. Kim

Writing, review, and/or revision of the manuscript: W.K. Kim, H. Kim

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.K. Kim, Y.J. Kim, N. Shin, H. Kim

Study supervision: W.K. Kim, H. Kim

This work was supported by the National Research Foundation of Korea grant funded by the Korean government [Ministry of Education, Science, and Technology (MEST); No. 2012R1A2A2A01005196], a grant of the Korean Health 21R&D Project, Ministry of Health and Welfare, Republic of Korea (A111218-CP01), Technology and by the Converging Research Center Program through the Ministry of Education, Science and Technology (2010K001115), and a grant from the Korea Healthcare Technology R&D Project, Ministry for Health & Welfare Affairs, Republic of Korea (A085136).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.